Music Emotion Recommender System using Spectral Features-a Malayalam cine music deployment
D. Vidyanadha Babu, P. Supriya · 2021 Second International Conference on Electronics and Sustainable Communication Systems (ICESC) · 2021
Songs tell a story with an interplay of different emotions. Hence, emotion analysis is widely deployed in today's music-based entertainment applications. Here, Malayalam cine music is analyzed for mood recommendation using spectral features and classification algorithms. Spectral features along with rhythm feature, tempo is extracted from the database created. The study attempts to compare the performance of models built with spectral features and rhythm feature(tempo). The extracted features are trained for classification using two algorithms- Support Vector Machines (SVM) and Artificial Neural Networks (ANN). Songs belonging to four categories namely- happy, calm, energetic and sad are considered here. The testing accuracy for simulation in case of neural networks is 80%. Performance of the models with a larger database is to be studied for future work.